Lecture 29 RCBD & Unequal Cell Sizes

Total Page:16

File Type:pdf, Size:1020Kb

Lecture 29 RCBD & Unequal Cell Sizes Lecture 29 RCBD & Unequal Cell Sizes STAT 512 Spring 2011 Background Reading KNNL: 21.1-21.6, Chapter 23 29-1 Topic Overview • Randomized Complete Block Designs (RCBD) • ANOVA with unequal sample sizes 29-2 RCBD • Randomized complete block designs are useful whenever the experimental units are non-homogeneous. • Grouping EU’s into “blocks” of homogeneous units helps reduce the SSE and increase the likelihood that we will be able to see differences among treatments. • A “block” consists of a complete replication of the set of treatments. Blocks and treatments are assumed not to interact. 29-3 RCBD Model • Assuming no replication, same as two-way ANOVA with one observation per cell. No interaction between block and treatment. Yijk=µρ + i + τ j + ε ijk iid where ε∼N 0, σ 2 and ρ= τ = 0 ijk ( ) ∑i ∑ i • We refer to ρi as the block effects and τj as the treatment effects. • We are really only interested in further analysis on the treatment effects. 29-4 RCBD Example • Want to study the effects of three different sealers on protecting concrete patios from the weather. • Ten unsealed patios are available spread across Indianapolis. • Separate each patio into three portions, and apply the treatments (randomly) in such a way that each patio receives each treatment for 1/3 of the surface. 29-5 RCBD Example (2) • Patio (location) is a blocking factor. Probably the weather will be different in each location; some patios may be better sheltered (trees, etc.) • If patio location is important, then failing to block on patio location would probably mean that the MSE will be overestimated. • Blocking requires DF (9 in this case), but usually if blocking variable is unimportant, the MSE with/without blocking will be about the same. 29-6 RCBD Example (3) Source DF SS MS F Value patio 9 900 100 9.0 sealer 2 100 50 5.0 Error 18 180 10 Total 29 1180 • If the ANOVA results are as above, then blocking is clearly important. If we do not block here... Source DF SS MS F Value sealer 2 100 50 1.25 Error 27 1080 40 29-7 RCBD Example (4) Source DF SS MS F Value patio 9 108 12 1.2 sealer 2 100 50 5.0 Error 18 180 10 Total 29 388 • If the ANOVA results are as above, then blocking doesn’t appear to have been as important. In this case if we fail to block... Source DF SS MS F Value sealer 2 100 50 4.69 Error 27 288 10.7 29-8 Big Picture • Failing to block when you should block can cost you the ability to see treatment effects • Blocking when there is no need usually often doesn’t cost much at all (though it can if the SSBlock is small enough relative to df). • Blocking effectively requires foresight. An experimenter must guess what sources of variation will exist in order to block on them. 29-9 Other Advantages of RCBD • Reasonably simple analysis to perform • Effective grouping makes results much more precise. • Can drop an entire block or treatment if necessary, without complicating the analysis. • Can deliberately introduce extra variability into the EU’s to widen the range of validity of the results without sacrificing precision. 29-10 Some Disadvantages of RCBD • Missing observations are a complex problem (since generally each treatment is represented exactly once in each block) • Loss of error degrees of freedom • Additional assumptions are required for the model (additivity, constant variance across blocks) 29-11 Multiple Blocking Variables • Often is the case that you the EU’s have multiple characteristics on which you could block. • Example: Consider the effect of three treatments for asthma. Might block on both AGE and GENDER. • Each treatment would be represented once at each AGE*GENDER combination. 29-12 More on Blocking... • Quite a bit more information in Chapter 21 o More than one replicate per block o Factorial treatments • Would discuss this and related topics in STAT 514. 29-13 Unequal Sample Sizes • Encountered for a variety of reasons including: Convenience – usually if we have an observational study, we have very little control over the cell sizes. Cost Effectiveness – sometimes the cost of samples is different, and we may use larger sample sizes when the cost is less In experimental studies, you may start with a balanced design, but lose that balance if problems occur. 29-14 Unequal Sample Sizes (2) • What changes? Loss of balance brings “intercorrelation” among the predictors (i.e., variables are no longer orthogonal) Type I and III SS will be different; typically Type III SS should be used for testing LSMeans should be used for testing Standard errors for cell means and for multiple comparisons will be different Confidence intervals will have different widths 29-15 Example • Examine the effects of gender (A) and bone development (B) on the rate of growth induced by a synthetic growth hormone. • Three categories of Bone Development Depression (Severe, Moderate, and Mild) • We categorize people on this basis after they are in the study (it is an observational factor); we wouldn’t want to throw away data just to keep a balanced design. • Page 954, growth.sas 29-16 Data / Sample Sizes Severe Moderate Mild 1.4 2.1 0.7 Male 2.4 1.7 1.1 2.2 2.4 2.5 0.5 Female 1.8 0.9 2.0 1.3 29-17 29-18 Interpretation • Same as any interaction plot • Effect seems to be greater if disease is more severe. • Effect seems greater for women than men. • Possibly an interaction. The effect of bone development is enhanced (greater) for women as compared to men. • We aren’t saying anything about significance here – we’ll do that when we look at the ANOVA. 29-19 ANOVA Output Source DF SS MS F Value Pr > F Model 5 4.474 0.895 5.51 0.0172 Error 8 1.300 0.163 Total 13 5.774 R-Square Root MSE growth Mean 0.774864 0.403113 1.642857 29-20 Type I / III SS Source DF Type I SS MS F Value Pr > F gender 1 0.00286 0.00286 0.02 0.8978 bone 2 4.39600 2.19800 13.53 0.0027 gen*bone 2 0.07543 0.03771 0.23 0.7980 Source DF Type III SS MS F Value Pr > F gender 1 0.1200 0.1200 0.74 0.4152 bone 2 4.1897 2.0949 12.89 0.0031 gen*bone 2 0.0754 0.0377 0.23 0.7980 29-21 Type X SS • There are actually four relevant types of sums of squares. I – Sequential II – Added Last (Observation) III – Added Last (Cell) IV – Added Last (Empty Cells) 29-22 Types I SS • Sequential Sums of Squares, appropriate for equal cell sizes. • SS(A), SS(B|A), SS(A*B|A,B) • Each observation is weighted equally , with the result that treatments are weighted in proportion to their cell size (if unequal, then not all treatments get the same weight in the analysis) 29-23 Types II SS • Variable Added Last SS, appropriate for equal cell sizes. • SS(A|B,A*B), SS(B|A,A*B), SS(A*B|A,B) • Each observation is weighted equally 29-24 Types III SS • Variable Added Last SS, appropriate for unequal cell sizes. • SS(A|B,A*B), SS(B|A,A*B), SS(A*B|A,B) • Each cell/treatment is weighted equally , but observations are weighted differently. Type III SS adjusts for the fact that cell sizes are different, unequal weighting of observations. 29-25 Type IV SS • Variable Added Last SS, necessary if there are empty cells • SS(A|B,A*B), SS(B|A,A*B), SS(A*B|A,B) • Like Type III SS but additionally takes into account the possibility of empty cells. 29-26 Data: Design Chart Severe Moderate Mild Male xxx xx xx Female x xxx xxx 29-27 Example: Type I Hypotheses Main Effect Gender 3 2 2 1 3 3 H 0: 7µ 11+ 7 µ 12 + 7 µ 13 = 7 µ 21 + 7 µ 22 + 7 µ 23 Main Effect Bone 3 1 2 3 2 3 H 0: 4µ 11+ 4 µ 21 =5 µ 12 + 5 µ 22 = 5 µ 13 + 5 µ 23 Observations weighted equally, treatment weighted by sample size. 29-28 Example: Type III Hypotheses Main Effect Gender 1 1 H 0: 3(µµµ 11++ 12 13) = 3 ( µµµ 21 ++ 22 23 ) Main Effect Bone 1 1 1 H 0: 2(µµ 11+= 21) 2( µµ 12 += 22) 2 ( µµ 13 + 23 ) Treatments are weighted equally, observations not weighted equally. 29-29 General Strategy • Remember that Type I SS and Type III SS examine different null hypotheses. • Type III SS are preferred when sample sizes are not equal, but can be somewhat misleading if sample sizes differ greatly. • Type IV SS are appropriate if there are empty cells. • Can obtain Type II/IV SS if necessary by using /ss1 ss2 ss3 ss4 in MODEL statement 29-30 Example: Type III SS Source DF Type III SS MS F Value Pr > F gender 1 0.1200 0.1200 0.74 0.4152 bone 2 4.1897 2.0949 12.89 0.0031 gen*bone 2 0.0754 0.0377 0.23 0.7980 • The interaction and gender effects are not significant. • Now look at comparing different levels of bone; should not ‘change’ models at this point, so need to average over gender. 29-31 Multiple Comparisons • Suppose we keep model as is, and examine effect of bone.
Recommended publications
  • When Does Blocking Help?
    page 1 When Does Blocking Help? Teacher Notes, Part I The purpose of blocking is frequently described as “reducing variability.” However, this phrase carries little meaning to most beginning students of statistics. This activity, consisting of three rounds of simulation, is designed to illustrate what reducing variability really means in this context. In fact, students should see that a better description than “reducing variability” might be “attributing variability”, or “reducing unexplained variability”. The activity can be completed in a single 90-minute class or two classes of at least 45 minutes. For shorter classes you may wish to extend the simulations over two days. It is important that students understand not only what to do but also why they do what they do. Background Here is the specific problem that will be addressed in this activity: A set of 24 dogs (6 of each of four breeds; 6 from each of four veterinary clinics) has been randomly selected from a population of dogs older than eight years of age whose owners have permitted their inclusion in a study. Each dog will be assigned to exactly one of three treatment groups. Group “Ca” will receive a dietary supplement of calcium, Group “Ex” will receive a dietary supplement of calcium and a daily exercise regimen, and Group “Co” will be a control group that receives no supplement to the ordinary diet and no additional exercise. All dogs will have a bone density evaluation at the beginning and end of the one-year study. (The bone density is measured in Houndsfield units by using a CT scan.) The goals of the study are to determine (i) whether there are different changes in bone density over the year of the study for the dogs in the three treatment groups; and if so, (ii) how much each treatment influences that change in bone density.
    [Show full text]
  • Lec 9: Blocking and Confounding for 2K Factorial Design
    Lec 9: Blocking and Confounding for 2k Factorial Design Ying Li December 2, 2011 Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design 2k factorial design Special case of the general factorial design; k factors, all at two levels The two levels are usually called low and high (they could be either quantitative or qualitative) Very widely used in industrial experimentation Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design Example Consider an investigation into the effect of the concentration of the reactant and the amount of the catalyst on the conversion in a chemical process. A: reactant concentration, 2 levels B: catalyst, 2 levels 3 replicates, 12 runs in total Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design 1 A B A = f[ab − b] + [a − (1)]g − − (1) = 28 + 25 + 27 = 80 2n + − a = 36 + 32 + 32 = 100 1 B = f[ab − a] + [b − (1)]g − + b = 18 + 19 + 23 = 60 2n + + ab = 31 + 30 + 29 = 90 1 AB = f[ab − b] − [a − (1)]g 2n Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design Manual Calculation 1 A = f[ab − b] + [a − (1)]g 2n ContrastA = ab + a − b − (1) Contrast SS = A A 4n Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design Regression Model For 22 × 1 experiment Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design Regression Model The least square estimates: The regression coefficient estimates are exactly half of the \usual" effect estimates Ying Li Lec 9: Blocking and Confounding for 2k Factorial Design Analysis Procedure for a Factorial Design Estimate factor effects.
    [Show full text]
  • Chapter 7 Blocking and Confounding Systems for Two-Level Factorials
    Chapter 7 Blocking and Confounding Systems for Two-Level Factorials &5² Design and Analysis of Experiments (Douglas C. Montgomery) hsuhl (NUK) DAE Chap. 7 1 / 28 Introduction Sometimes, it is impossible to perform all 2k factorial experiments under homogeneous condition. I a batch of raw material: not large enough for the required runs Blocking technique: making the treatments are equally effective across many situation hsuhl (NUK) DAE Chap. 7 2 / 28 Blocking a Replicated 2k Factorial Design 2k factorial design, n replicates Example 7.1: chemical process experiment 22 factorial design: A-concentration; B-catalyst 4 trials; 3 replicates hsuhl (NUK) DAE Chap. 7 3 / 28 Blocking a Replicated 2k Factorial Design (cont.) n replicates a block: each set of nonhomogeneous conditions each replicate is run in one of the blocks 3 2 2 X Bi y··· SSBlocks= − (2 d:f :) 4 12 i=1 = 6:50 The block effect is small. hsuhl (NUK) DAE Chap. 7 4 / 28 Confounding Confounding(干W;混雜;ø絡) the block size is smaller than the number of treatment combinations impossible to perform a complete replicate of a factorial design in one block confounding: a design technique for arranging a complete factorial experiment in blocks causes information about certain treatment effects(high-order interactions) to be indistinguishable(p|辨½的) from, or confounded with blocks hsuhl (NUK) DAE Chap. 7 5 / 28 Confounding the 2k Factorial Design in Two Blocks a single replicate of 22 design two batches of raw material are required 2 factors with 2 blocks hsuhl (NUK) DAE Chap. 7 6 / 28 Confounding the 2k Factorial Design in Two Blocks (cont.) 1 A = 2 [ab + a − b−(1)] 1 (any difference between block 1 and 2 will cancel out) B = 2 [ab + b − a−(1)] 1 AB = [ab+(1) − a − b] 2 (block effect and AB interaction are identical; confounded with blocks) hsuhl (NUK) DAE Chap.
    [Show full text]
  • Introduction to Biostatistics
    Introduction to Biostatistics Jie Yang, Ph.D. Associate Professor Department of Family, Population and Preventive Medicine Director Biostatistical Consulting Core In collaboration with Clinical Translational Science Center (CTSC) and the Biostatistics and Bioinformatics Shared Resource (BB-SR), Stony Brook Cancer Center (SBCC). OUTLINE What is Biostatistics What does a biostatistician do • Experiment design, clinical trial design • Descriptive and Inferential analysis • Result interpretation What you should bring while consulting with a biostatistician WHAT IS BIOSTATISTICS • The science of biostatistics encompasses the design of biological/clinical experiments the collection, summarization, and analysis of data from those experiments the interpretation of, and inference from, the results How to Lie with Statistics (1954) by Darrell Huff. http://www.youtube.com/watch?v=PbODigCZqL8 GOAL OF STATISTICS Sampling POPULATION Probability SAMPLE Theory Descriptive Descriptive Statistics Statistics Inference Population Sample Parameters: Inferential Statistics Statistics: 흁, 흈, 흅… 푿 , 풔, 풑 ,… PROPERTIES OF A “GOOD” SAMPLE • Adequate sample size (statistical power) • Random selection (representative) Sampling Techniques: 1.Simple random sampling 2.Stratified sampling 3.Systematic sampling 4.Cluster sampling 5.Convenience sampling STUDY DESIGN EXPERIEMENT DESIGN Completely Randomized Design (CRD) - Randomly assign the experiment units to the treatments Design with Blocking – dealing with nuisance factor which has some effect on the response, but of no interest to the experimenter; Without blocking, large unexplained error leads to less detection power. 1. Randomized Complete Block Design (RCBD) - One single blocking factor 2. Latin Square 3. Cross over Design Design (two (each subject=blocking factor) 4. Balanced Incomplete blocking factor) Block Design EXPERIMENT DESIGN Factorial Design: similar to randomized block design, but allowing to test the interaction between two treatment effects.
    [Show full text]
  • Design of Engineering Experiments Blocking & Confounding in the 2K
    Design of Engineering Experiments Blocking & Confounding in the 2 k • Text reference, Chapter 7 • Blocking is a technique for dealing with controllable nuisance variables • Two cases are considered – Replicated designs – Unreplicated designs Chapter 7 Design & Analysis of Experiments 1 8E 2012 Montgomery Chapter 7 Design & Analysis of Experiments 2 8E 2012 Montgomery Blocking a Replicated Design • This is the same scenario discussed previously in Chapter 5 • If there are n replicates of the design, then each replicate is a block • Each replicate is run in one of the blocks (time periods, batches of raw material, etc.) • Runs within the block are randomized Chapter 7 Design & Analysis of Experiments 3 8E 2012 Montgomery Blocking a Replicated Design Consider the example from Section 6-2 (next slide); k = 2 factors, n = 3 replicates This is the “usual” method for calculating a block 3 B2 y 2 sum of squares =i − ... SS Blocks ∑ i=1 4 12 = 6.50 Chapter 7 Design & Analysis of Experiments 4 8E 2012 Montgomery 6-2: The Simplest Case: The 22 Chemical Process Example (1) (a) (b) (ab) A = reactant concentration, B = catalyst amount, y = recovery ANOVA for the Blocked Design Page 305 Chapter 7 Design & Analysis of Experiments 6 8E 2012 Montgomery Confounding in Blocks • Confounding is a design technique for arranging a complete factorial experiment in blocks, where the block size is smaller than the number of treatment combinations in one replicate. • Now consider the unreplicated case • Clearly the previous discussion does not apply, since there
    [Show full text]
  • Biostatistics and Experimental Design Spring 2014
    Bio 206 Biostatistics and Experimental Design Spring 2014 COURSE DESCRIPTION Statistics is a science that involves collecting, organizing, summarizing, analyzing, and presenting numerical data. Scientists use statistics to discern patterns in natural systems and to predict how those systems will react in different situations. This course is designed to encourage an understanding and appreciation of the role of experimentation, hypothesis testing, and data analysis in the sciences. It will emphasize principles of experimental design, methods of data collection, exploratory data analysis, and the use of graphical and statistical tools commonly used by scientists to analyze data. The primary goals of this course are to help students understand how and why scientists use statistics, to provide students with the knowledge necessary to critically evaluate statistical claims, and to develop skills that students need to utilize statistical methods in their own studies. INSTRUCTOR Dr. Ann Throckmorton, Professor of Biology Office: 311 Hoyt Science Center Phone: 724-946-7209 e-mail: [email protected] Home Page: www.westminster.edu/staff/athrock Office hours: Monday 11:30 - 12:30 Wednesday 9:20 - 10:20 Thursday 12:40 - 2:00 or by appointment LECTURE 11:00 – 12:30, Tuesday/Thursday Patterson Computer Lab Attendance in lecture is expected but you will not be graded on attendance except indirectly through your grades for participation, exams, quizzes, and assignments. Because your success in this course is strongly dependent on your presence in class and your participation you should make an effort to be present at all class sessions. If you know ahead of time that you will be absent you may be able to make arrangements to attend the other section of the course.
    [Show full text]
  • Running Head: EXPERIMENTAL and QUASI-EXPERIMENTAL DESIGNS 1
    Running Head: EXPERIMENTAL AND QUASI-EXPERIMENTAL DESIGNS 1 EXPERIMENTAL AND QUASI-EXPERIMENTAL DESIGNS IN VISITOR STUDIES: A CRITICAL REFLECTION ON THREE PROJECTS Scott Pattison, TERC Josh Gutwill, Exploratorium Ryan Auster, Museum of Science, Boston Mac Cannady, Laurence Hall of Science This is the pre-publiCation version of the following artiCle: Pattison, S., Gutwill, J., Auster, R., & Cannady, M. (2019). Experimental and quasi-experimental designs in visitor studies: A critical reflection on three projects. Visitor Studies, 22(1), 43–66. https://doi.org/10.1080/10645578.2019.1605235 1 of 42 Running Head: EXPERIMENTAL AND QUASI-EXPERIMENTAL DESIGNS 2 Abstract Identifying causal relationships is an important aspect of research and evaluation in visitor studies, such as making claims about the learning outcomes of a program or exhibit. Experimental and quasi-experimental approaches are powerful tools for addressing these causal questions. However, these designs are arguably underutilized in visitor studies. In this article, we offer examples of the use of experimental and quasi-experimental designs in science museums to aide investigators interested in expanding their methods toolkit and increasing their ability to make strong causal claims about programmatic experiences or relationships among variables. Using three designs from recent research (fully randomized experiment, post-test only quasi- experimental design with comparison condition, and post-test with independent pre-test design), we discuss challenges and trade-offs related
    [Show full text]
  • Randomized Complete Block Designs
    1 2 RANDOMIZED COMPLETE BLOCK DESIGNS Randomization can in principal be used to take into account factors that can be treated by blocking, but Introduction to Blocking blocking usually results in smaller error variance, hence better estimates of effect. Thus blocking is Nuisance factor: A factor that probably has an effect sometimes referred to as a method of variance on the response, but is not a factor that we are reduction design. interested in. The intuitive idea: Run in parallel a bunch of Types of nuisance factors and how to deal with them experiments on groups (called blocks) of units that in designing an experiment: are fairly similar. Characteristics Examples How to treat The simplest block design: The randomized complete Unknown, Experimenter or Randomization block design (RCBD) uncontrollable subject bias, order Blinding of treatments v treatments Known, IQ, weight, Analysis of (They could be treatment combinations.) uncontrollable, previous learning, Covariance measurable temperature b blocks, each with v units Known, Temperature, Blocking Blocks chosen so that units within a block are moderately location, time, alike (or at least similar) and units in controllable (by batch, particular different blocks are substantially different. choosing rather machine or (Thus the total number of experimental units than adjusting) operator, age, is n = bv.) gender, order, IQ, weight The v experimental units within each block are randomly assigned to the v treatments. (So each treatment is assigned one unit per block.) 3 4 Note that experimental units are assigned randomly RCBD Model: only within each block, not overall. Thus this is sometimes called a restricted randomization.
    [Show full text]
  • Design of Engineering Experiments the Blocking Principle
    Design of Engineering Experiments The Blocking Principle • Montgomery text Reference, Chapter 4 • Bloc king and nuiftisance factors • The randomized complete block design or the RCBD • Extension of the ANOVA to the RCBD • Other blocking scenarios…Latin square designs 1 The Blockinggp Principle • Blocking is a technique for dealing with nuisance factors • A nuisance factor is a factor that probably has some effect on the response, but it’s of no interest to the experimenter…however, the variability it transmits to the response needs to be minimized • Typical nuisance factors include batches of raw material, operators, pieces of test equipment, time (shifts, days, etc.), different experimental units • Many industrial experiments involve blocking (or should) • Failure to block is a common flaw in designing an experiment (consequences?) 2 The Blocking Principle • If the nuisance variable is known and controllable, we use blocking • If the nuisance factor is known and uncontrollable, sometimes we can use the analysis of covariance (see Chapter 15) to remove the effect of the nuisance factor from the analysis • If the nuisance factor is unknown and uncontrollable (a “lurking” variable), we hope that randomization balances out its impact across the experiment • Sometimes several sources of variability are combined in a block, so the block becomes an aggregate variable 3 The Hardness Testinggp Example • Text reference, pg 120 • We wish to determine whether 4 different tippps produce different (mean) hardness reading on a Rockwell hardness tester
    [Show full text]
  • Learning Blocking Schemes for Record Linkage∗
    Learning Blocking Schemes for Record Linkage∗ Matthew Michelson and Craig A. Knoblock University of Southern California Information Sciences Institute, 4676 Admiralty Way Marina del Rey, CA 90292 USA {michelso,knoblock}@isi.edu Abstract Record linkage is the process of matching records across data sets that refer to the same entity. One issue within record linkage is determining which record pairs to consider, since a detailed comparison between all of the records is imprac- tical. Blocking addresses this issue by generating candidate matches as a preprocessing step for record linkage. For ex- ample, in a person matching problem, blocking might return all people with the same last name as candidate matches. Two main problems in blocking are the selection of attributes for generating the candidate matches and deciding which meth- ods to use to compare the selected attributes. These attribute and method choices constitute a blocking scheme. Previ- ous approaches to record linkage address the blocking issue in a largely ad-hoc fashion. This paper presents a machine learning approach to automatically learn effective blocking Figure 1: Record linkage example schemes. We validate our approach with experiments that show our learned blocking schemes outperform the ad-hoc blocking schemes of non-experts and perform comparably to or cluster the data sets by the attribute. Then we apply the those manually built by a domain expert. comparison method to only a single member of a block. Af- ter blocking, the candidate matches are examined in detail Introduction to discover true matches. This paper focuses on blocking. Record linkage is the process of matching records between There are two main goals of blocking.
    [Show full text]
  • Experimental and Quasi-Experimental Designs for Research
    CHAPTER 5 Experimental and Quasi-Experimental Designs for Research l DONALD T. CAMPBELL Northwestern University JULIAN C. STANLEY Johns Hopkins University In this chapter we shall examine the validity (1960), Ferguson (1959), Johnson (1949), of 16 experimental designs against 12 com­ Johnson and Jackson (1959), Lindquist mon threats to valid inference. By experi­ (1953), McNemar (1962), and Winer ment we refer to that portion of research in (1962). (Also see Stanley, 19S7b.) which variables are manipulated and their effects upon other variables observed. It is well to distinguish the particular role of this PROBLEM AND chapter. It is not a chapter on experimental BACKGROUND design in the Fisher (1925, 1935) tradition, in which an experimenter having complete McCall as a Model mastery can schedule treatments and meas~ In 1923, W. A. McCall published a book urements for optimal statistical efficiency, entitled How to Experiment in Education. with complexity of design emerging only The present chapter aspires to achieve an up­ from that goal of efficiency. Insofar as the to-date representation of the interests and designs discussed in the present chapter be­ considerations of that book, and for this rea­ come complex, it is because of the intransi­ son will begin with an appreciation of it. gency of the environment: because, that is, In his preface McCall said: "There afe ex­ of the experimenter's lack of complete con­ cellent books and courses of instruction deal­ trol. While contact is made with the Fisher ing with the statistical manipulation of ex; tradition at several points, the exposition of perimental data, but there is little help to be that tradition is appropriately left to full­ found on the methods of securing adequate length presentations, such as the books by and proper data to which to apply statis­ Brownlee (1960), Cox (1958), Edwards tical procedure." This sentence remains true enough today to serve as the leitmotif of 1 The preparation of this chapter bas been supported this presentation also.
    [Show full text]
  • Randomized Complete Block Design
    Randomized Complete Block Designs Randomized complete block designs differ from the completely randomized designs in that the experimental units are grouped into blocks according to known or suspected variation which is isolated by the blocks. Variation such as fertility, sand, and wind gradients, or age and litter of animals can be isolated by appropriate blocking. Therefore, within each block, the conditions are as homogeneous as possible, but between blocks, large differences may exist. In the attached figure 4 blocks have been established which subdivide the gradient into smaller segments. This results in relatively small gradients within each block so that the treatments may be compared under relatively homogeneous conditions. The treatments are assigned within the individual blocks at random with a separate randomization for each block. The analysis of variance table indicates that the error term now has 9 df rather than 12 df as in the CR design. These 3 df were removed for the block effect which removed the variation due to the gradient from the error variance. This will result in a more precise test of the treatment effects since the mean square for error will be smaller and the F value for treatment should be larger. Advantages of randomized complete block designs 1. Complete flexibility. Can have any number of treatments and blocks. 2. Provides more accurate results than the completely randomized design due to grouping. 3. Relatively easy statistical analysis even with missing data. 4. Allows calculation of unbiased error for specific treatments. Disadvantages of randomized complete block designs 1. Not suitable for large numbers of treatments because blocks become too large.
    [Show full text]